Rheumatoid arthritis: the predictable effect of small immune complexes in which antibody is also antigen.

T existence of autoantibodies and immune (ii) both the synovial and extra-articular features of RA are the predictable outcome of activation of complexes in rheumatoid arthritis (RA) has been recognized for several decades. Antibodies to IgG, or macrophages bearing the receptor FccRIIIa by immune complexes of this type, with the producrheumatoid factors (RF), were reported nearly 60 yr ago [1, 2]. A pathogenic role for immune complexes, tion of cytokines such as TNF-a, and resulting inflammation; and those formed by RF of the IgG class in particular, seemed likely in the 1970s, but has become less fashion(iii) in synovium, TNF-a also induces changes in fibroblasts which facilitate the local accumulation able, being championed only by a few, notably Roitt, Mannik and Natvig [3–6 ]. Interest in RF-based of ectopic lymphoid tissue and plasma cells, with the generation of high local levels of RF and the immune complexes waned for three reasons. Firstly, there is no obvious reason why these should specifically formation of larger immune complex aggregates which induce further inflammatory events, includproduce joint disease. Secondly, the relationship between RF and disease is not one to one. Thirdly, ing complement activation. the putative pathogenic effects of immune complexes Perhaps the most important aspect of this hypothesis were considered chiefly in terms of complement is that it refocuses attention on the possibility that activation. By 1990, cytokines such as tumour necrosis permanent interruption of autoantibody production factor alpha (TNF-a) had come to be considered the might effectively cure the disease. We propose that this more important mediators in the joint, and therapeutic might be possible. trials using cytokine blockade have reinforced that Much of the hypothesis is based on conventional view [7]. immunological dogma. Although large IgM-based As early as 1983, it had been shown that immune immune complexes may form in the circulation of RA, complexes, including those derived from RF, could these should bind complement and be cleared via stimulate macrophages to produce soluble factors (i.e. complement receptors. As reviewed by Davies [14], cytokines) capable of inducing extracellular matrix large complexes may cause problems in systemic lupus degradation [8]. It seemed likely that the complexes primarily because the complement system is defective. were binding to IgG Fc receptors (FccR), but not Kunkel and colleagues identified smaller complexes in until 10 yr later had work by Fanger, Lydyard, the circulation in RA, chiefly in the form of IgG RF Huizinga, van de Winkel and others led to a clear self-associating dimers [4]. These small complexes fix understanding of the structure and function of these complement very poorly [15]. IgG dimers are also receptors [9–11]. Moreover, only recently has it been small enough to be expected to gain significant access established that synovial intima is a highly specialized to the extravascular space and allow interaction with immunological microenvironment which includes an tissue macrophages. unusual pattern of FccR expression [12, 13]. Three classes of FccR on macrophages might bind Armed with this new information, we are now in a IgG RF complexes [10]. FccRI is expressed at rather position to construct a hypothesis which may resolve low levels on macrophages (~104/cell ) unless stimuthe difficulties perceived in the 1980s and point again lated by, for example, c-interferon. Interestingly, FccRI to a central role for immune complexes, and specifically is expressed very weakly by synovial intimal cells in for IgG RF, in the pathogenesis of RA. RA. It has a high affinity for free IgG and may be We propose that: saturated with monomeric IgG under most conditions. Binding of soluble IgG complexes has chiefly been (i) the self-associated dimeric complexes of IgG RF attributed to the other two receptor classes. FccRII found in the blood of RA subjects are small is present at higher levels on all macrophages enough both to evade clearance by complement (~105/cell ), but is of low affinity and there is evidence receptors and to cross endothelium to reach the from granulocytes, at least, that binding of very small extravascular space; (dimeric) complexes to FccRII alone will not activate the cell, and that the presence of FccRIII is required Submitted 18 August 1997; revised version accepted as well. FccRIII is of moderate affinity, will bind 24 November 1997. dimeric complexes and binding of the FccRIIIa isoform Correspondence to: J. C. W. Edwards, Rheumatology Unit, Arthur Stanley House, 40–50 Tottenham Street, London W1P 9PG. present on macrophages and natural killer (NK ) cells